Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A computer-implemented method for processing a search query, comprising: receiving the search query from a user; associating the search query with a statement of an entry from a ground truth table, when the entry is in the ground truth table; adding the query as the statement to the ground truth table when the statement is not in the ground truth table; retrieving a set of search results from the ground truth table; presenting the set of search results, with a restatement of the search query, including a stock keeping unit number (SKU) and a numerical ranking associated with the SKU; recording an interaction of the user with the set of search results; and modifying the ground truth table based on the interaction of the user; wherein it is determined that the statement is not in the ground truth table based on a returned value from the search query being below a predetermined threshold.
This invention relates to a computer-implemented method for improving search query processing by dynamically updating a ground truth table to enhance search result accuracy. The method addresses the problem of inconsistent or low-quality search results by leveraging user interactions to refine and expand a reference dataset. The system receives a search query from a user and checks if the query matches any entry in a ground truth table, which stores verified statements and associated search results. If the query does not match an existing entry, it is added to the table as a new statement. The system then retrieves search results from the table, presenting them alongside the query, including product identifiers (SKU numbers) and numerical rankings. User interactions with these results—such as clicks or selections—are recorded and used to modify the ground truth table, improving future search accuracy. The determination of whether a query matches an existing entry is based on a threshold value derived from the search query's returned results. This approach ensures that the ground truth table evolves with user behavior, dynamically adapting to provide more relevant and reliable search outcomes. The method is particularly useful in e-commerce or information retrieval systems where search quality directly impacts user experience and engagement.
2. The method of claim 1 , wherein the modifying the ground truth table includes adjusting the numerical ranking within an entry of the ground truth table.
This invention relates to improving data processing systems that use ground truth tables for training or validation purposes. The problem addressed is the static nature of ground truth tables, which may not accurately reflect real-world variations or evolving data relationships. The invention provides a method to dynamically modify a ground truth table by adjusting the numerical rankings within its entries. This adjustment can involve changing the order of ranked items, altering their relative scores, or updating their positions based on new data or performance metrics. The modification process ensures that the ground truth table remains relevant and accurate over time, improving the reliability of systems that depend on it. The method may also include validating the modified table to confirm that the adjustments maintain or enhance its utility for training or validation tasks. This approach is particularly useful in applications where data relationships change frequently, such as recommendation systems, search engines, or machine learning models that require up-to-date reference data. By dynamically adjusting the ground truth table, the system can adapt to new information without requiring a complete rebuild of the reference data.
3. The method of claim 1 , wherein recording an interaction of the user comprises recording a click-through action.
A system and method for tracking user interactions with digital content involves capturing and analyzing user engagement data to improve content delivery and user experience. The technology addresses the challenge of understanding how users interact with digital interfaces, such as websites or applications, to optimize design, content placement, and user engagement strategies. The method records various user interactions, including click-through actions, which involve tracking when a user selects or clicks on a specific element within the digital interface. This data is collected to analyze user behavior, identify popular or frequently accessed content, and refine the presentation of information. The recorded interactions may also include other types of engagement, such as scrolling, hovering, or time spent on specific sections, to provide a comprehensive view of user activity. By capturing these interactions, the system enables content providers to make data-driven decisions, such as adjusting layout, improving navigation, or personalizing content based on user preferences. The recorded data can be processed to generate insights, such as heatmaps or engagement metrics, which help in optimizing the user experience and increasing conversion rates. The method ensures that user interactions are accurately logged and analyzed to enhance the effectiveness of digital content delivery.
4. The method of claim 1 , wherein recording an interaction of the user comprises recording a purchase made by the user of a related item to the set of search results.
This invention relates to user interaction tracking in search systems, specifically for improving search result relevance by analyzing user behavior. The problem addressed is the difficulty in determining which search results are most useful to users based solely on traditional metrics like click-through rates. The invention enhances search systems by recording detailed user interactions, including purchases of items related to the search results, to refine future search queries and rankings. The method involves monitoring user actions after a search is performed. When a user interacts with a search result, such as clicking on a link or making a purchase, this interaction is recorded. The recorded interactions are then analyzed to identify patterns, such as which search results lead to purchases of related items. This data is used to adjust the ranking of search results, prioritizing those that have a higher likelihood of leading to user engagement or conversions. The system may also track additional interactions, such as time spent on a page or subsequent searches, to further refine relevance. By incorporating purchase data as a key interaction metric, the system improves the accuracy of search result rankings, ensuring that users are more likely to find and purchase relevant items. This approach enhances user satisfaction and increases the effectiveness of search engines in e-commerce and other domains where user intent and behavior are critical.
5. The method of claim 1 , wherein recording an interaction of the user comprises recording a wish list addition action.
A system and method for tracking user interactions with digital content, particularly in e-commerce or content recommendation platforms, addresses the challenge of capturing and analyzing user preferences to improve personalized recommendations. The method involves monitoring and recording user actions, such as adding items to a wish list, to infer user interests. When a user adds an item to a wish list, the system captures this interaction as a data point, which is then processed to refine recommendation algorithms. The recorded interactions may include additional metadata, such as timestamps, item details, or user context, to enhance the accuracy of preference modeling. By analyzing these interactions, the system can identify patterns in user behavior, predict future preferences, and tailor content or product suggestions accordingly. This approach improves the relevance of recommendations, increasing user engagement and satisfaction. The method may also integrate with other user activity tracking mechanisms, such as browsing history or purchase behavior, to build a comprehensive profile of user preferences. The system ensures data privacy by anonymizing or encrypting user-specific information where necessary. The recorded wish list additions serve as explicit indicators of user intent, distinguishing them from passive interactions like browsing, thereby providing more reliable signals for recommendation engines. This method enhances the efficiency of personalized content delivery in digital platforms.
6. The method of claim 1 , wherein recording an interaction of the user comprises recording a favorite list addition action.
A system and method for tracking user interactions with digital content, particularly focusing on user preferences and engagement. The technology addresses the challenge of capturing and analyzing user behavior to improve content recommendations, personalization, and user experience. The method involves monitoring and recording various user actions, including the specific action of adding items to a favorite list. This action is logged as part of a broader interaction dataset, which may include other user activities such as clicks, views, or selections. The recorded interactions are then processed to derive insights about user preferences, which can be used to enhance content delivery, tailor recommendations, or refine user interfaces. The system may integrate with content platforms, social media, or digital media services to provide a seamless and personalized experience. By capturing favorite list additions, the method helps identify content that resonates with users, enabling more accurate and relevant suggestions. The recorded data can also be used for analytics, trend analysis, or behavioral studies to optimize user engagement and satisfaction. The approach ensures that user interactions are tracked in a structured manner, facilitating efficient data processing and actionable insights.
7. The method of claim 1 , wherein recording an interaction of the user comprises recording a share action.
A system and method for tracking user interactions with digital content, particularly focusing on sharing actions. The technology addresses the need to monitor how users engage with content, including sharing it across platforms or with other users. The method involves capturing detailed data about user interactions, specifically recording when a user performs a share action, such as distributing content via social media, email, or messaging platforms. This data is then processed to analyze sharing patterns, user behavior, and content popularity. The system may also track additional interaction types, such as likes, comments, or saves, to provide a comprehensive view of user engagement. The recorded data can be used for analytics, content optimization, or targeted advertising. The method ensures accurate and detailed tracking of sharing actions, enabling businesses and content creators to better understand how their content is disseminated and engaged with by users. This helps in refining content strategies and improving user experience based on real-world interaction data.
8. The method of claim 1 , wherein recording an interaction of the user comprises recording a duration spent by the user on the set of search results.
A system and method for analyzing user interactions with search results to improve search relevance. The technology addresses the problem of inefficient search result ranking by tracking how users engage with displayed results, allowing search engines to refine future queries based on observed behavior patterns. The method involves presenting a set of search results to a user, recording interactions with those results, and using the interaction data to adjust the ranking or relevance of future search results. Interaction recording includes capturing the duration a user spends viewing each result, which indicates the perceived relevance of that result. Additional interaction data may include clicks, dwell time on specific results, and navigation patterns. The system processes this data to identify trends, such as frequently clicked or ignored results, and updates the search algorithm accordingly. By continuously refining rankings based on real user behavior, the system improves search accuracy and user satisfaction. The method applies to web search engines, enterprise search systems, and any application where search result relevance is critical. The recorded interaction data may be stored and analyzed to further optimize search performance over time.
9. The method of claim 1 , wherein recording an interaction of the user comprises recording a number of accesses by the user of the set of search results.
This invention relates to user interaction tracking in search systems, specifically improving the analysis of how users engage with search results. The problem addressed is the lack of detailed data on user behavior when interacting with search results, which limits the ability to refine search algorithms and user experience. The method involves recording and analyzing user interactions with a set of search results. Specifically, it tracks the number of times a user accesses individual search results, providing quantitative data on which results are most frequently engaged with. This tracking can be combined with other interaction metrics, such as dwell time or clicks, to build a comprehensive profile of user behavior. The recorded data is then used to improve search result relevance, personalize future searches, or optimize the presentation of results. The system may include a search engine that generates a set of search results in response to a user query. A tracking module records the user's interactions, including the frequency of accesses to each result. This data is stored and analyzed to identify patterns, such as which results are most or least accessed. The analysis can be used to adjust ranking algorithms, prioritize certain results, or refine the search system's performance over time. The method ensures that user behavior directly influences the search system's evolution, enhancing accuracy and relevance.
10. The method of claim 2 , wherein the interaction includes a click-through action, a purchase made by the user of a related item to the set of search results, a wish list addition action, a favorite list addition action, a share action, a duration spent by the user on the set of search results, and a number of accesses by the user of the set of search results.
This invention relates to improving search result relevance in digital systems by analyzing user interactions with search results. The problem addressed is the inefficiency of traditional search algorithms that do not adequately incorporate user behavior to refine future search outcomes. The solution involves tracking and processing various user interactions with search results to enhance the accuracy and personalization of subsequent searches. The method involves monitoring specific user actions related to search results, including click-through actions, purchases of related items, additions to wish lists or favorite lists, sharing actions, the duration spent viewing results, and the frequency of accessing the results. These interactions are used to evaluate the relevance of the search results and adjust the search algorithm accordingly. By analyzing these behaviors, the system can better understand user preferences and intent, leading to more accurate and tailored search results in future queries. The approach improves upon prior systems by considering a broader range of user interactions beyond simple clicks, providing a more comprehensive understanding of user engagement. This enhances the overall search experience by delivering more relevant and personalized results, increasing user satisfaction and efficiency in finding desired information or products. The method is applicable to various digital platforms, including e-commerce, content search engines, and recommendation systems.
11. The method of claim 2 , wherein the adjusting a ranking within an entry of the ground truth table comprises increasing or decreasing the ranking of the entry.
A system and method for adjusting rankings within a ground truth table to improve data accuracy and reliability. The ground truth table stores reference data used for training or evaluating machine learning models, where each entry has an associated ranking indicating its reliability or relevance. The method involves dynamically modifying these rankings by either increasing or decreasing the ranking of specific entries based on performance metrics, feedback, or other criteria. This adjustment ensures that higher-ranked entries are prioritized in training or evaluation processes, while lower-ranked entries are deprioritized or removed. The system may also include mechanisms to validate the adjusted rankings, ensuring consistency and accuracy. By dynamically updating rankings, the method enhances the quality of the ground truth data, leading to improved model performance and reliability. The approach is particularly useful in applications where reference data quality directly impacts model accuracy, such as in natural language processing, computer vision, or other AI-driven fields. The method may be integrated into larger data management or machine learning pipelines to automate ranking adjustments and maintain high-quality reference datasets.
12. The method of claim 1 , wherein modifying the ground truth table includes creating a new entry.
13. A system for processing a search query, comprising: a search query server, the search query server comprising: a processor; a memory coupled to the processor, wherein the memory contains instructions, that when executed by the processor, perform the steps of: receiving the search query from a user; associating the search query with a statement of an entry from a ground truth table, when the entry is in the ground truth table; adding the query as the statement to the ground truth table when the statement is not in the ground truth table; retrieving a set of search results from the ground truth table; presenting the set of search results, with a restatement of the search query, including a stock keeping unit number (SKU) and a numerical ranking associated with the SKU; recording an interaction of the user with the set of search results; and modifying the ground truth table based on the recorded interaction of the user; wherein it is determined that the statement is not in the ground truth table based on a returned value from the search query being below a predetermined threshold.
The system processes search queries by leveraging a ground truth table to improve search result accuracy and relevance. The system includes a search query server with a processor and memory containing instructions for executing the following steps. When a user submits a search query, the system checks if the query matches any entry in the ground truth table. If a match is found, the query is associated with the corresponding statement. If no match is found, the query is added as a new entry to the ground truth table, provided the search query returns a result below a predetermined threshold. The system then retrieves search results from the ground truth table and presents them to the user, including a restatement of the query, a stock keeping unit (SKU) number, and a numerical ranking for each SKU. User interactions with the search results are recorded, and the ground truth table is updated based on these interactions to refine future search results. This approach ensures that the search system continuously learns from user behavior to enhance accuracy and relevance.
14. The system of claim 13 , wherein the memory further contains instructions, that when executed by the processor, perform the step of adjusting the numerical ranking within an entry of the ground truth table.
The invention relates to a system for managing and adjusting numerical rankings within a ground truth table used in machine learning or data processing applications. The ground truth table stores reference data or labels that serve as the correct answers for training or evaluating machine learning models. The system includes a processor and a memory containing instructions that, when executed, perform various operations related to the ground truth table. One key function is adjusting the numerical ranking within an entry of the ground truth table. This adjustment may involve modifying the priority, weight, or order of data entries to improve the accuracy or performance of the system. The system may also include a user interface for inputting or modifying the ground truth data, as well as a display for visualizing the rankings or adjustments. The adjustments can be based on user input, automated algorithms, or feedback from the machine learning model. The overall goal is to enhance the reliability and effectiveness of the ground truth table in training or evaluating machine learning models.
15. The system of claim 13 , wherein the memory further contains instructions, that when executed by the processor, perform the steps of recording a click-through action of the user, and wherein modifying the ground truth table is based on the click-through action of the user.
This invention relates to an adaptive user interface system that improves content recommendations by dynamically updating a ground truth table based on user interactions. The system addresses the problem of static recommendation models that fail to adapt to changing user preferences over time. The ground truth table stores reference data used to train or refine recommendation algorithms, ensuring they align with actual user behavior. The system includes a processor and memory storing instructions for tracking user interactions, such as clicks or selections, and modifying the ground truth table accordingly. When a user clicks on a recommended item, this action is recorded, and the ground truth table is updated to reflect the user's preference. This feedback loop allows the system to continuously refine its recommendations, improving accuracy and relevance. The system may also include a display for presenting recommendations and an input device for capturing user interactions. By dynamically adjusting the ground truth table based on real-time user feedback, the system ensures that recommendations remain aligned with evolving user interests. This adaptive approach enhances user satisfaction and engagement by delivering more personalized content. The invention is particularly useful in digital advertising, content platforms, and recommendation engines where user preferences are dynamic.
16. The system of claim 13 , wherein the memory further contains instructions, that when executed by the processor, perform the steps of recording a purchase action of the user, and wherein modifying the ground truth table is based on the purchase action of the user.
This invention relates to a system for dynamically updating a ground truth table in a recommendation system based on user purchase actions. The system operates in the domain of personalized recommendations, where the challenge is to accurately reflect user preferences in real-time to improve recommendation accuracy. The ground truth table stores verified user preferences, which are used to train or refine recommendation algorithms. The system includes a processor and memory containing instructions for monitoring user interactions, such as clicks or views, to infer preferences and update the ground truth table accordingly. Additionally, the system records user purchase actions, which serve as strong indicators of actual user preferences. The ground truth table is then modified based on these purchase actions, ensuring that the most reliable signals (purchases) override weaker signals (e.g., clicks or views). This dynamic updating mechanism improves the accuracy of future recommendations by prioritizing verified user behavior over inferred preferences. The system may also include a user interface for displaying recommendations and a network interface for communicating with external data sources. The overall goal is to enhance recommendation systems by continuously refining the ground truth table with high-confidence user actions, leading to more personalized and relevant suggestions.
17. The system of claim 13 , wherein the memory further contains instructions, that when executed by the processor, perform the steps of recording a wish list addition action of the user, and wherein modifying the ground truth table is based on the wish list addition action of the user.
This invention relates to a system for personalizing recommendations based on user behavior, particularly wish list additions. The system addresses the challenge of improving recommendation accuracy by dynamically updating a ground truth table—a data structure storing verified user preferences—to reflect actions like adding items to a wish list. The system includes a processor and memory storing instructions for tracking user interactions, such as wish list additions, and modifying the ground truth table accordingly. By incorporating these actions, the system refines its understanding of user preferences, leading to more relevant recommendations. The ground truth table serves as a reference for generating personalized suggestions, ensuring that the system adapts to evolving user interests over time. This approach enhances recommendation quality by leveraging explicit user signals, such as wish list additions, to update the underlying preference model. The system may also include additional components for processing user data, such as a data processing module to analyze interactions and a recommendation engine to generate tailored suggestions. The overall goal is to create a more responsive and accurate recommendation system that aligns with user preferences as they change.
18. The system of claim 13 , wherein the memory further contains instructions, that when executed by the processor, perform the steps of recording a favorite list addition action of the user, and wherein modifying the ground truth table is based on the favorite list addition action of the user.
This invention relates to a system for personalizing content recommendations based on user interactions, particularly focusing on how user actions like adding items to a favorites list influence recommendation algorithms. The system includes a processor and memory storing instructions that, when executed, enable the system to track and analyze user behavior to refine content suggestions. Specifically, the system records when a user adds an item to a favorites list and uses this action to update a ground truth table, which serves as a reference for generating recommendations. The ground truth table is dynamically modified based on these user actions, ensuring that the recommendations align with the user's preferences over time. This approach improves the accuracy and relevance of personalized content suggestions by incorporating explicit user feedback, such as favorites, into the recommendation process. The system may also include other features, such as tracking user interactions with recommended content and adjusting recommendations based on engagement metrics. The overall goal is to enhance user satisfaction by delivering more tailored and relevant content recommendations.
19. The system of claim 13 , wherein the memory further contains instructions, that when executed by the processor, perform the steps of recording a share action of the user, and wherein modifying the ground truth table is based on the share action of the user.
This invention relates to a system for dynamically updating a ground truth table based on user interactions, particularly in the context of machine learning or data analysis. The system includes a processor and a memory storing instructions that, when executed, perform several functions. The system receives input data and processes it to generate output data, which is then compared against a ground truth table to determine accuracy. The system also records user interactions, such as sharing or annotating data, and uses these interactions to modify the ground truth table. This ensures the ground truth table remains accurate and relevant over time, adapting to new data patterns or user feedback. The system may also include a user interface for displaying the output data and receiving user input, as well as a communication interface for transmitting data to external systems. The dynamic updating mechanism helps improve the reliability of the system's outputs by incorporating real-world user behavior into the ground truth reference. This is particularly useful in applications where data distributions or user preferences evolve, such as recommendation systems, content moderation, or predictive analytics.
20. A computer program product for processing a search query on an electronic device, comprising a computer readable hardware storage device having program instructions embodied therewith, the program instructions executable by a processor to cause the electronic device to: receive the search query from a user; associate the search query with a statement of an entry from a ground truth table, when the entry is in the ground truth table; add the query as the statement to the ground truth table when the statement is not in the ground truth table; retrieve a set of search results from the ground truth table; present the set of search results, with a restatement of the search query, including a stock keeping unit number (SKU) and a numerical ranking associated with the SKU; record an interaction of the user with the set of search results; and modify the ground truth table based on the interaction of the user; wherein it is determined that the statement is not in the ground truth table based on a returned value from the search query being below a predetermined threshold.
This invention relates to improving search query processing on electronic devices, particularly for e-commerce or inventory-based systems where accurate and relevant search results are critical. The problem addressed is the inefficiency of traditional search systems that fail to dynamically adapt to user interactions, leading to suboptimal search results and user frustration. The system receives a search query from a user and checks if the query matches any existing entries in a ground truth table, which stores predefined statements and their associated search results. If the query matches an entry, the system retrieves and presents the corresponding search results, including a restatement of the query, a stock keeping unit (SKU) number, and a numerical ranking for each SKU. If the query does not match any entry, it is added to the ground truth table as a new statement. The system then retrieves and presents search results for the new query. User interactions with the results, such as clicks or selections, are recorded and used to modify the ground truth table, improving future search accuracy. The determination of whether a query matches an existing entry is based on whether the returned search results fall below a predetermined threshold, indicating low relevance or confidence. This adaptive approach ensures that the search system continuously learns from user behavior to provide more accurate and personalized results over time.
Unknown
April 7, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.